The notion that physicians can be socially incentivized by the presence of their peers is intuitive. However, the nature of these "peer effects" among physicians has not previously been studied. The objective of this project is to study the effect of social networks on the behavior of individual physicians and on the overall productivity of health care organizations. Specifically, we will evaluate the effects of emergency department physicians on each other as they work together. Specific aims are threefold. First, we will measure peer effects on individual productivity - along dimensions of speed of care, quality of care, and teaching. We predict tradeoffs between more observable dimensions (e.g., speed of care) and less observable ones (e.g., quality of care), and we expect that peer effects will be stronger among more socially connected peers. Second, we will study peer effects on the choice of patients by physicians. Individual physicians may want to choose easy patients (i.e., "cherry pick"), but social networks may mitigate this by social pressure and appropriate coordination of work. Third, we will study the effect on overall emergency department productivity under a new system that disallows physicians from choosing their own patients. Preliminary results show that emergency physicians differ in their productivity and that they often have strong reported preferences for certain types of patients, which they may share with or differ from their peers. Given this, we will use two complementary features of our research design to accomplish our specific aims. First, we have detailed, longitudinal data on individual physicians and the patients that they see, involving sufficient variation in the teams of physicians working with each other. Second, the emergency department being studied is undertaking a natural experiment in May of 2011 in which the rules of patient assignment will disallow physicians from choosing their own patients. This research will shed light on how health care organizations should design their social networks implicit in team structures and the rules that govern behavior within teams. Specific implications include how organizations should match peers with each other, how transparent they should make productivity measures among peers at the expense of less transparent measures, and how much they should constrain implicitly social behaviors such as the choice of patients among physicians. PUBLIC HEALTH RELEVANCE: Like any other type of worker, physicians may be influenced in their behavior by the presence of their colleagues or "peers." This research project studies such "peer effects" on productivity and choice of patients. In addition to studying the behavior of physicians as they work among different colleagues, this project will study the overall impact on peer effects in a natural experiment in which the same emergency physicians are no longer allowed to choose their own patients.